scientific correspondence
distinct phenomena. Antimalarial antibod-
ies defined by agglutination, enzyme-linked
immunosorbent assay (data not shown) or
surface immunofluorescence (data not
shown) were not associated with the pres-
ence or absence of infection.
Anti-adhesion antibodies against CSA-
binding parasites differ from antibodies
against parasites with other binding pheno-
types. In studies of CD36-binding parasites,
sera were variably effective4–6 and blocked
adhesion strain-specifically; the inhibition
of adhesion may have been associated with
agglutination, which can interfere with
adhesion. Our data indicate that anti-
adhesion antibodies against CSA-binding
parasites uniformly develop in multigravid
women from holoendemic areas, appear in
Asian and African women, block adhesion
in a strain-independent manner, and are
distinct from agglutinating antibodies.
We have shown that women with anti-
adhesion antibodies against CSA-binding
parasites are protected from malaria infec-
tion during pregnancy. Because women
have limited exposure to the CSA-binding
parasite before first gestation, primigravidas
lack anti-adhesion antibodies against this
parasite subpopulation. We propose that
this absence of anti-adhesion antibodies
increases susceptibility to malaria during
first pregnancies. Further research is
required to examine the association
between anti-adhesion antibodies and preg-
nancy outcome, and, most important, to
identify PfCSA-L as the next step towards
developing a vaccine. An anti-adhesion vac-
cine that conferred protection against
maternal malaria would benefit millions of
pregnant women and infants in the tropics,
and should be a public health priority.
Michal Fried*, François Nosten†,
Alan Brockman†, Bernard J. Brabin‡,
Patrick E. Duffy*
*Department of Immunology,
Walter Reed Army Institute of Research,
14th and Dahlia Street,
Washington DC 20307, USA
e-mail: friedm@wrsmtp-ccmail.army.mil
†Shoklo Malaria Research Unit,
PO Box 46, Mae Sot, 63110, Thailand;
Faculty of Tropical Medicine,
Mahidol University, Bangkok, Thailand;
and Centre for Tropical Medicine,
Nuffield Department of Medicine,
John Radcliffe Hospital, Oxford OX3 9DU, UK
‡Tropical Child Health Group,
Liverpool School of Tropical Medicine,
Pembroke Place, Liverpool L3 5QA, UK
1. McGregor, I. A. Am. J. Trop. Med. Hyg. 33, 517–525 (1984).
2. Fried, M. & Duffy, P. E. Science 272, 1502–1504 (1996).
3. Nosten, F., ter Kuile, F., Maelankirri, L., Decludt, B. & White, N. J.
Trans. R. Soc. Trop. Med. Hyg. 85, 424–429 (1991).
4. Singh, B. et al. Clin. Exp. Immunol. 72, 145–150 (1988).
5. Southwell, B. R. et al. Trans. R. Soc. Trop. Med. Hyg. 83, 464–469
(1989).
6. Udeinya, I. J., Miller, L. H., McGregor, I. A. & Jensen, J. B.
Nature 303, 429–431 (1983).
852
Object recognition can
drive motion perception
When two spatially separated spots of light
are flashed in rapid succession, apparent
motion is seen between them1–3. We exten-
ded this phenomenon by photographing a
face and producing from it a fragmented
‘puzzle picture’ or ‘Mooney face’ in which
the face is not initially visible (Fig. 1, left;
frame 1) but is seen after 15 to 60 seconds4.
Another photograph of the same face seen
in profile was used to produce a second
Mooney face (Fig. 1, right; frame 2). When
the two images were alternated, naive sub-
jects at first saw random, incoherent, two-
dimensional (2D) motion between the
fragments. But once the face was recog-
nized, it was perceived to rotate unambigu-
ously in three dimensions. We conclude
that complex image tokens set up by per-
ceptual learning can drive perception of
apparent motion.
To what extent are different aspects of
perception mediated by specialized mod-
ules5–8 in the brain? Attention can influence
the way in which corresponding points are
seen to move3 and, in ambiguous, bi-stable
displays, people tend to see ‘plausible’ tra-
jectories (for example, a bomb is more likely
to be seen going down rather than up9). But
it is usually assumed that these are weak
second-order effects, no more than a small
bias in the perceived direction of motion.
Cells in area MT of the monkey brain
respond to apparent motion10 and there is
psychophysical evidence that primitive,
rather than sophisticated, visual features are
used for motion correspondence11. As a
result, most computational models of
motion perception emphasize a stronger
modular approach12,13, arguing that motion
correspondence is not influenced by high-
level object knowledge and semantics. For
instance, it is easy to see a pig’s face trans-
forming into a human face even though this
is logically absurd.
Nature © Macmillan Publishers Ltd 1998
Here we use a new type of apparent-
motion display to investigate directly how
perceptual learning and high-level object
knowledge are involved in motion process-
ing. First, we illuminated the face of a
model looking at the camera lens with a
8
single strong light source. We then pho-
tographed the face and digitized the image
to obtain a fragmented puzzle picture (Fig.
1, left). The model then rotated his face by
45 degrees (or 90 degrees) and a second
image was made using the same procedure
(Fig. 1, right). A set of four pairs of such
images was made of four different faces.
The direction of illumination and degree of
head rotation were chosen to eliminate local
matches between the shadows and to opti-
mize the difference between the appearance
of the images before and after perceptual
learning4.
Naive subjects could often not see the
faces in these images. Of 28 subjects, 17
could not see the face initially in at least one
of the four sets of photographs, and four of
these subjects could not do so in two. These
17 volunteers were first shown various
demonstrations of apparent motion (but
not of faces) to familiarize them with the
illusion. They were then shown frames 1
and 2 of the images in which no face was
perceived, alternated at 3 Hz (ISI 5 0) for
20 seconds, and asked to describe the per-
ceived direction of motion, if any. The sub-
jects saw either random chaotic motion or
motion along arbitrary 2D trajectories that
often varied from trial to trial (in 20 of 21
experimental sessions). When asked explic-
itly if they saw any 3D motion, 16 subjects
said they had not; the seventeenth subject
immediately saw a face rotate in one of the
displays as soon as the two frames were
alternated.
The film sequence was then stopped and
the same subjects were asked whether they
could see a human face, a prompt that
always resulted in the face being seen after
10 to 30 seconds (we often had to resort
to pointing to individual features after 30
seconds). Remarkably, when the same two
frames were alternated, all subjects in all
Figure 1 Two views of a
‘Mooney’-type face pro-
duced by digitizing a
video image. Left, frame
1; right, frame 2. A
demonstration can be
produced by showing the
two frames alternately as
an upside-down movie;
nearly all subjects see 2D
movement. But if the
same movie is viewed
upright or the subjects
are told it is a face, they
invariably see rigid 3D
rotation.
NATURE | VOL 395 | 29 OCTOBER 1998 | www.nature.com

trials saw 3D rotation of the face, from
frontal to semiprofile or profile. No 2D
expansion, contraction or chaotic incoher-
ent motion was ever seen. Because the dis-
play was exactly the same but the perceived
motion trajectory had changed completely,
scientific correspondence
and examined under a light microscope,
which revealed that all organisms died at
Preserving tardigrades pressures over 200 Mpa (Fig. 1).
We then investigated whether tardi-
under pressure grades could acquire pressure resistance in
the tun state (a process known as anhydro-

we conclude that high-level object represen-
tations can drive apparent motion. In par-
ticular, a temporary ‘object’ (or extended
contours forming portions of it) created
exclusively by perceptual learning can pro-
vide an input for apparent motion. The
meaning of the perceived configuration
therefore helps to eliminate an infinite set of
false matches between the fragments, allow-
ing motion correspondence to be estab-
lished correctly between the two views of
the face.
Similar effects were seen with other
puzzle pictures, such as a hidden dalmatian
dog14, with the dog sitting and standing in
alternate frames. When the two frames were
alternated, subjects initially saw random 2D
motion, but once the dog was seen it was
perceived to sit and stand alternately. This
effect is particularly remarkable because the
second image was made using a completely
different dog with different spots, so no
rigid transformation was possible between
the two images.
Taken together, these observations sug-
gest that strong interactions occur between
perceptual modules concerned with motion
(for example, in area V5)5 and those
involved in high-level object recognition
(for example, in area IT)4. The strictly hier-
archical, modular models of perception
popularized by computer scientists should
therefore be replaced by a more dynamic
view. From every stage in the hierarchy, par-
tial answers may be sent back to bias the
processing in earlier stages.
V. S. Ramachandran, C. Armel, C. Foster,
R. Stoddard
Center for Brain and Cognition,
University of California at San Diego,
La Jolla, California 92093, USA
When an animal is exposed to high hydro-
static pressure, its cellular membranes1,2,
proteins3–5 and DNA6 are damaged. At
pressures of around 30 megapascals (MPa),
proliferation and metabolism in micro-
organisms stops; at 300 MPa, most bacteria
and multicellular organisms die. But here
we show that, in perfluorocarbon at pres-
sures as high as 600 MPa, small terrestrial
animals known as tardigrades can survive
in a dehydrated state.
Terrestrial tardigrades become immobile
and shrink into a form known as the ‘tun’
state when the humidity decreases. In this
state, they can survive extreme tempera-
tures, as low as 2253 °C or as high as 151
°C, as well as exposure to a vacuum or to
X-rays 7–10. We have now tested the ability of
the tardigrades Macrobiotus occidentalis
(order Eutardigrada) and Echiniscus japoni-
cus (Heterotardigrada) to survive under
extraordinarily high hydrostatic pressures.
We sealed M. occidentalis tardigrades in a
small plastic container (6 ml) placed inside a
pressure capsule (R7K-3-10, Yamamoto
Suiatu Kougyousho) and compressed using
water as the pressure medium. The outside
temperature was 21 °C and the water tem-
perature inside the capsule was 25 °C;
stepped hydrostatic pressures were applied
for 20 minutes at a time (100, 200, 300, 400,
500 and 600 MPa). Pressure was increased
by 100 MPa per minute and then decreased
at the same rate. After decompression, M.
occidentalis was removed from the capsule
100
***
90
80
Survival rate (%)
8
biosis) by dehydrating them before apply-
ing pressure. M. occidentalis and E.
japonicus were dehydrated on filter paper in
Petri dishes for more than 24 hours, when
the relative humidity in the dishes dropped
from 70–80% to 10–30%. To prevent the
tardigrades from rehydrating during com-
pression, we used an inert solvent, perfluo-
rocarbon (C8F18 Fluorinate PC77,
Sumitomo 3M), as the pressure medium
instead of water. Tardigrades were then
removed from the pressure capsule and
soaked in water to rinse off the perfluoro-
carbon. One hour later, we confirmed that
they had changed from the tun state to the
active state.
To test whether the perfluorocarbon
increased the survival of tardigrades
exposed to high hydrostatic pressure, we
subjected tardigrades in perfluorocarbon,
which were still in the active state, to the
same hydrostatic pressure changes. All
active-state tardigrades were dead at pres-
sures above 200 MPa.
We evaluated the data at 600 MPa for the
group (n 5 20) in the tun state in perfluoro-
carbon and in the active state in water
and perfluorocarbon. The survival rate of
M. occidentalis was 95% at 600 MPa (Fig.1),
and there was a difference between active-
state and tun-state animals (P , 0.01;
chi-squared). The survival rate of E. japoni-
cus was 80%, as some animals had died
and their fluid had leaked onto the filter
paper, which we attributed to inadequate
***
*** ***
***

70 *** P < 0.01

e-mail: vramacha@ucsd.edu
60
tun state in
1. Helmhotz, H. L. F. von Handbuch der Psysiologischen Optik
50 perfluorocarbon
(Leipzig, 1867).
2. Kolers, P. Aspects of Motion Perception (Pergamon, New York, 40 active state in
water
1972). 30 active state in
3. Ramachandran, V. S. & Anstis, S. M. Sci. Am. 254, 102–109 perfluorocarbon
(1986). 20
4. Tovee, M. J., Rolls, E. T. & Ramachandran, V. S. Neuroreport 7, 10
2757–2760 (1996).
0
5. Zeki, S. M. Nature 274, 423–428 (1978).
6. Allman, J. in Oxford Companion to the Mind (ed. Gregory, R. L.) 0 100 200 300 400 500 600
(Oxford Univ. Press, 1987). Pressure (MPa)
7. Livingstone, M. S. & Hubel, D. H. J. Neurosci. 7, 3416–3468
(1987). Figure 1 Survival rate of Macrobiotus occidentalis after exposure to high hydrostatic pressure. Before com-
8. Van Essen, D. C. Annu. Rev. Neurosci. 2, 227–263 (1979). pression (100 MPa min21), tardigrades were either in the tun state (dehydrated) in perfluorocarbon, or in the
9. Bruner, J. Psychol. Rev. 64, 123–152 (1957).
active state in water or perfluorocarbon. Either water or perfluorocarbon was used as pressure medium. An
10. Mikami, A., Newsome, W. T. & Wurtz, R. H. J. Neurophysiol. 55,
1308–1327 (1986). hour after high hydrostatic pressure was maintained for 20 minutes and decompression to 2100 MPa min21,
11. Ramachandran, V. S. & Inada, V. Spatial Vis. 1, 57–67 (1985). the animals were soaked in distilled water for an hour, and those in the active state were examined with an
12. Marr, D. Vision (Freedom, San Francisco, 1982). optical microscope (magnification, 3 40). When animals in the tun state were exposed to 100–600 MPa in
13. Ullman, S. The Interpretation of Visual Motion (MIT Press,
perfluorocarbon, a high survival rate was obtained. The survival rate of active-state animals in perfluorocar-
Cambridge, MA, 1979).
14. Gregory, R. L. The Intelligent Eye (McGraw Hill, New York, bon is 80% at 100 MPa and 0% at pressures greater than 200 MPa. This survival rate at 100 MPa was lower
1970). than the 91% obtained using water at 100 MPa, so perfluorocarbon did not increase the survival rate.

NATURE | VOL 395 | 29 OCTOBER 1998 | www.nature.com Nature © Macmillan Publishers Ltd 1998 853